- Email: [email protected]
Acidity distribution of carboxyl groups in Loy Yang brown coal: its analysis and the change by heat treatment
Journal of Colloid and Interface Science 260 (2003) 176–183 www.elsevier.com/locate/jcis
Acidity distribution of carboxyl groups in Loy Yang brown coal: its analysis and the change by heat treatment Kenji Murakami,∗ Ryuhei Kondo, Kiyoshi Fuda, and Toshiaki Matsunaga Department of Materials-Process Engineering & Applied Chemistry for Environments, Faculty of Engineering and Resource Science, Akita University, Akita 010-8502, Japan Received 30 July 2002; accepted 24 November 2002
Abstract Brown coals have a considerable number of acidic functional groups of which the main component is carboxyl groups, and the acidity has a wide distribution. In this paper, changes of the acidity distribution were examined by aqueous titration when brown coal was heat-treated to control its acidity distribution. For Loy Yang brown coal from Australia dried at 50 ◦ C under vacuum (LY50), the acid dissociation constant, Ka, was distributed over a wide pKa range between 2 and 9. Then, using Gaussian functions, the acidity distribution was divided into four groups, which were characterized by average pKa values: average pKa value of 3.8 (hereafter referred to as Group A), 5.2 (Group B), 6.8 (Group C), and 8.3 (Group D). Among them, Groups A, B, and C were assigned to carboxyl groups. From the changes of the number of carboxyl groups when brown coal was heat-treated up to 400 ◦ C, it was found that the way of decrease was different among these acidic groups. The decrease of the amount of carboxyl groups in Group C was significant, and at 325 ◦ C most of them disappeared. On the other hand, the carboxyl groups in Group A remained even at a high temperature of 400 ◦ C. We estimated approximately the structures around carboxyl groups for LY50 and their structural changes by heat treatment using the known pKa values for simple carboxylic acids and the pKa values calculated by the MOPAC program for complicated carboxylic acids. 2003 Elsevier Science (USA). All rights reserved. Keywords: Brown coal; Carboxyl group; Acidity distribution; Heat treatment; Titration; MOPAC calculation
1. Introduction It is well known that brown coals have the ability of cation exchange and carboxyl groups play an important role [1–13]. Individual carboxyl group existing in brown coal should have different acidity due to the variety of environmental conditions, including the substrates bound with carboxyl groups. In the field of catalytic study, analytical methods such as NH3 -TPD and IR study of adsorbed amines have been used to investigate the acidity of solid acids. However, the surface acidity measured by such solid–gas reactions should be considerably different from that measured by solid–liquid reactions (e.g., the adsorption of metallic ions onto solid acids in aqueous solutions). We examined the acidity distribution of carboxyl groups on the surface of Australian Loy Yang brown coal from the results of aqueous titration in our earlier report [11]. However, at that time we * Corresponding author.
E-mail address: [email protected] (K. Murakami).
obtained only a small number of experimental data; therefore we could obtain only rough acid-dissociating properties of carboxyl groups. Recently, by analyzing a proton-binding isotherm (this isotherm represents the change of the protonbinding number of acidic functional groups as a function of the pH of the solution obtained by aqueous titration, and this term is used by Schwarz et al.), the surface acidities of carbonaceous materials such as activated carbons [14–21] and other materials such as titania [22], mixed oxides (TiO2 –SiO2 , ZrO2 –SiO2 ) [23], and carbon composite adsorbents (oxidized carbon and titanium silicate) [24] have been evaluated. According to these reports, it was clear that there were some acidic functional groups with different acidities on the surfaces of these heterogeneous materials. In this study, we made automatic titration equipment which can drop small amounts of reagent to increase the number of experimental data and to analyze the titration data in detail. Here, we assumed that the acidity distribution function was Gaussian, similarly to our earlier study [11] and Contescu et al. [20]. As described later, we confirmed that a very sharp acidity distribution could be obtained by ana-
0021-9797/03/$ – see front matter 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0021-9797(02)00172-8
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
lyzing the acidity distributions of some pure carboxylic acids (from monobasic acid to tribasic acid). The purpose of this study is to clarify the acidity distribution of brown coal and to investigate the change in acidity distribution when brown coal was heat-treated to control its acidity distribution. Also, we estimated the structural changes around carboxyl groups by comparing the pKa values calculated by the MOPAC program for carboxylic acids with unknown pKa values.
2. Experimental 2.1. Sample preparation Loy Yang brown coal from Victoria, Australia was used in this study. Raw coal was ground below particle size 250 µm, washed with deionized water, dried at 50 ◦ C under vacuum, and stored in a desiccator. This sample is a starting material (hereafter referred to as LY50). The analyses for LY50 are as follows: C 67.6% (daf), H 5.2% (daf), N 0.8% (daf), O 26.4% (diff), and ash 0.2% (dry). A sample of 10 g of LY50 was heat-treated at a rate of 10 ◦ C min−1 up to desired temperatures (250–400 ◦ C) under nitrogen flow (50 ml min−1 ) and then held at the temperatures for 2 h (LY250–LY400). To examine the reliability of the experimental and analytical method used in this study, some carboxylic acids (acetic acid, phthalic acid, and citric acid) which are characterized by known acid dissociation constants were used as standard reagents. 2.2. Titration method A weighed amount of sample (0.5 g) was added into a 200-ml Erlenmeyer flask containing 100 ml of 0.1 mol l−1 NaOH solution. The suspension was continuously saturated with nitrogen to eliminate the influence of atmospheric CO2 and stirred magnetically throughout the measurements. The adsorption system was set in a vessel thermostated at 25 ± 0.2 ◦ C. The titrations were carried out with a 0.1 mol l−1 HCl solution using hand-made titration equipment and a pH meter (Horiba F-22) connected to a PC computer, which controlled the equilibrium condition and titrant delivery. The titration equipment is able to dose the titrant in increments of 1 ml to the suspension of samples. The pH of the solution was measured every 10 min. We assumed that the suspension reached equilibrium when the pH variation was less than both 0.01 pH units/10 min and 0.005 pH units/30 min.
expressed as Ka =
[COO− ][H+ ] , [COOH]
In an equilibrium condition, the acid dissociation constant, Ka, of weak acid such as carboxylic acids is generally
(1)
where [COOH], [COO− ], and [H+ ] are the molar concentrations (mol l−1 ) of undissociated carboxyl groups, dissociated carboxyl groups, and protons, respectively. Although fundamentally the acid dissociation constant should use activities instead of molar concentrations, we approximate the acid dissociation constant in this way. As described below, the results of simulations using this approximation for some carboxylic acids with known pKa values were in good correspondence with the values reported in the reference. Therefore, this approximation is considered to be valid. The degree of electrolytic dissociation, α, of carboxyl groups can be expressed in terms of Ka and [H+ ] as α=
[COO− ] Ka [COO− ] , (2) = = [COOH]T [COOH] + [COO− ] [H+ ] + Ka
where [COOH]T is the total molar concentration (mol l−1 ) of carboxylic acid. By differentiating Eq. (2) with respect to pH of the solution, the following equation can be obtained and is named q(pH, pKa): q(pH, pKa) =
loge 10 [H+]Ka . ([H+ ] + Ka)2
(3)
This Eq. (3) represents the number of dissociated carboxyl groups at each pH of the solution. In the case of a heterogeneous solid such as brown coal, the acid dissociation constant of carboxyl groups present on the surface is considered to have a wide distribution. This distribution being named as f (pKa), the dissociated amount of carboxyl groups at each pH, Q(pH), can be written in the form ∞ Q(pH) =
f (pKa)q(pH, pKa) dpKa,
(4)
−∞
where Q(pH) is a measurable amount. By deconvoluting Eq. (4), the acidity distribution, f (pKa), can be obtained. Although we do not know what function should be applied to the acidity distribution function, as described below we could reproduce the experimental values accurately by using a Gaussian function, which is most suitable for probability phenomena: f (pKa) = G(Ni , pKai , σi ) i
=
i
2.3. Analytical method of acidity distribution
177
pH − pKai exp −0.5 √ σi 2π σi Ni
2 .
(5)
In this study, f (pKa) was taken as an overlap of several Gaussian functions with the number of carboxyl groups Ni , the acid dissociation constant pKai , and the standard deviation σi .
178
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
3. Results and discussion 3.1. Examination of reliability of this experimental method by measuring the amount and acidity of carboxyl groups of standard reagents We evaluated the reliability of the experimental and analytical methods used in this study by titrating some pure materials such as acetic acid (monobasic acid), phthalic acid (dibasic acid), and citric acid (tribasic acid) as standard reagents. Figure 1 shows the procedure of analysis in the case of acetic acid as a example. The difference in sodium contents in solution between the blank titration curve and the titration curve with sample at the same pH, x in Fig. 1a, was taken as the amount of sodium binding. The curve showing the number of carboxyl groups bound with sodium ions as a function of the pH of the solution (Fig. 1b) was named the sodium binding isotherm. This isotherm was regarded as a measure of the amount of dissociated carboxyl groups. By differentiating this isotherm, the curve shown in Fig. 1c was
obtained. This differential curve is equivalent to Eq. (3) and named as the acid dissociation curve, Q(pH). The numerical values in the vertical axes (Figs. 1b and 1c) are normalized by the initial concentration of acetic acid. By deconvoluting Eq. (4) using Q(pH) obtained in the experiment, the f (pKa) of acetic acid was obtained (Fig. 1d). Similarly, for phthalic acid and citric acid, the sodium binding isotherms, the acid dissociation curves, and f (pKa) are shown in Figs. 2 and 3. Because we use a Gaussian function for analysis, the pKa value was distributed even in the case of carboxylic acids which should have a single pKa value inherently. But the widths of the acidity distribution were very narrow, as shown in Figs. 1d, 2c, and 3c (σi = 0.05–0.07 pKa units for any samples). For citric acid, these widths were narrower than the widths reported by other researchers, who analyzed by SAIEUS (the solution method for the integral equation using a B-spline function) [16] and CONTIN (a general purpose constrained regularization method) [24]. The analytical results are summarized in Table 1 together with the reference values [25]. All Ni parameters of each standard reagent are
Fig. 1. Procedure for calculation of acidity distribution for acetic acid. (a) Broken line is a blank curve and solid line is a titration curve for acetic acid. x indicates the amount of sodium binding. (b) The sodium binding isotherm. (c) The acid dissociation curve. Symbols, experimental data; solid line, calculated using Eq. (4). (d) The acidity distribution.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Fig. 2. (a) Sodium-binding isotherm for phthalic acid. (b) The acid dissociation curve for phthalic acid. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for phthalic acid.
179
Fig. 3. (a) Sodium-binding isotherm for citric acid. (b) The acid dissociation curve for citric acid. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for citric acid.
180
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Table 1 Calculated Ni , total Ni , and pKai parameters for standard reagents and the pKai values reported in Ref. [25] Ni (mol mol−1 ) Acetic acid 0.99 Phthalic acid 0.91 1.05 Citric acid 0.96 0.91 1.08
Total Ni (mol mol−1 )
pKai measured
pKai∗
0.99
4.62
4.56
1.96
2.74 4.98
2.75 4.93
2.95
3.04 4.38 5.66
2.90 4.34 5.66
pKai∗ : values reported in Ref. [25] (25 ◦ C, ionic strength = 0.1 mol dm−3 ).
about 1 mol l−1 , which are almost equal to their stoichiometric amounts. Further, the pKai parameters of all reagents are also in good agreement with the values reported in the reference [25]. Therefore, it is concluded that this method in our study can be used not only for quantitative but also for qualitative analysis of the titration results. 3.2. Calculation of acidity distribution and estimation of the structure around carboxyl groups for LY50 Figures 4a and 4b show the sodium-binding isotherm and the acid dissociation curve for LY50, respectively. This isotherm was an average of several experiments and the experimental values were in error by up to 4%. From Fig. 4b, the acid dissociation curve for LY50 was found to be very broad. In general, the acid dissociation constants of the carboxyl groups on the surface of carbonaceous materials such as brown coals and activated carbons are well known to have a wide distribution in the pKa range between 3 and 8. The reason the acid dissociation constants are distributed over such a wide range is considered to be the variety of interactions between carboxyl groups and substrates. The results of deconvolution of Eq. (4) are shown in Fig. 4c and Table 2. The obtained parameters (Ni , pKai , σi ) were further refined by fitting to the original sodium binding isotherm. Here, we used 10 Gaussian functions to deconvolute Eq. (4), but it was found from Fig. 4c that LY50 had only four groups of acidic functional groups in the pKa range between 2 and 9. The groups of acidic functional groups are referred to as Groups A, B, C, and D in order of increasing pKa values. The σi values for LY50 (σi = 0.35–0.36 pKa units) were about 5–7 times larger than those for carboxylic acids shown in Figs. 1d, 2c, and 3c, indicating clearly that the acid dissociation constants for LY50 have a wide distribution. We consider that heterogeneous materials with complicated structures such as brown coals, have a continuous pKa distribution over a much broader pKa range, but it was found from Fig. 4c that brown coal was composed of four groups of acidic functional groups which have relatively uniform chemical properties represented at these pKa values.
Fig. 4. (a) Sodium-binding isotherm for LY50. (b) The acid dissociation curve for LY50. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for LY50.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
181
Table 2 Calculated Ni , pKai , and total Ni parameters for coal samples Group A LY50 LY250 LY300 LY325 LY350 LY400
Group B
Group C
Ni
pKai
Ni
pKai
Ni
pKai
0.85 0.67 0.40 0.34 0.23 0.20
3.81 3.79 3.45 3.42 3.44 3.37
0.92 0.71 0.58 0.32 0.15 0
5.19 5.40 4.94 5.33 5.10 –
0.72 0.45 0.24 0 0 0
6.83 6.25 6.29 – – –
Total Ni 2.49 1.83 1.22 0.66 0.38 0.20
What kinds of acidic functional groups are represented at these four pKa values? According to Ref. [25], the acid dissociation constants of aliphatic monocarboxylic acids and benzoic acids are 4.5–4.9 and 4.0–4.6, respectively. On the other hand, the first and second acid dissociation constants of dicarboxylic acids are 1.7–4.5 and 4.1–5.8, respectively. These simple carboxylic acids themselves cannot be present in brown coal, but these pKa values may be useful for obtaining rough information on the structures around carboxyl groups. The average pKa value of Group A is about 3.8. Since this is a strong group, this acidic site would correspond with a structure similar to the benzoic acids and dicarboxylic acids (pK1 ). The average pKa value of Group B is about 5.2, which possibly corresponds with the pKa values of aliphatic monocarboxylic acids and the pK2 values of dicarboxylic acids. The average pKa value of Group C is about 6.8. For these carboxyl groups, we must consider the third acid dissociation constants where many carboxyl groups are present locally. The average pKa value of Group D is about 8.3, which is much higher than the pKa values of the simple carboxylic acids [25]. However, when we carried out the titration experiment for the weakly acidic cation-exchange resin Amberlite IRC-50, its acid dissociation constant was distributed over a wide pKa range, between 4 and 10. Consequently, there is a possibility that this Group D is also assigned to the carboxyl groups. On the other hand, Puziy et al. reported that the acidic site with pKa 7.7–8.0 might be assigned to phenolic hydroxyl groups [24]. In any case, since the acidic group with such higher pKa values is not actually concerned with the ion-exchange reaction, Group D is omitted from the following discussion. 3.3. Influence of heat treatment on the acidity distribution of carboxyl groups The sodium binding isotherms for the heat-treated coals are shown in Fig. 5. The acidity distribution functions for heat-treated coals were calculated similarly to the case of LY50, and results are shown in Fig. 6 and Table 2. For all samples, 10 Gaussian functions were used in deconvolution, but only four peaks appeared at the analogous pH values for LY50. Figure 7 shows the dependence of the number of carboxyl groups for each group on the heat-treatment temperature. From previous studies [26,27], it was clear that CO2 was produced by the decomposition of carboxyl
Fig. 5. Sodium-binding isotherms for heat-treated brown coals.
groups; therefore the number of carboxyl groups decreased with heat treatment. However, the way of decrease in number was different among acidic groups. Group A, which is the strongest acidic group, was difficult to decompose and 0.2 mmol g−1 of carboxyl groups remained, even at 400 ◦ C. On the other hand, the number of Group C decreased significantly and most of them disappeared at 325 ◦ C. Also, most of the carboxyl groups which belonged to Group B disappeared at 400 ◦ C. The average pKa values of each group are shown in Table 2. The average pKa value of Group A shifted from 3.8 to 3.4 gradually with increasing heat-treatment temperature. The average pKa value of Group B was not changed regularly and scattered at the center of about 5.2. From these results, we estimated the structural changes around carboxyl groups with heat treatment. From the fact that Group B and Group C were easy to decompose and that Group A was difficult, the more strongly acidic carboxyl groups were considered to remain without decomposition. But it is not necessarily concluded that only this decomposition behavior occurs, and there may be another possibility, that the pKa values of carboxyl groups which belonged to Group B or Group C shift to the pKa values of Group A with heat treatment. This can occur if carbonyl or ether groups are formed by dehydration or decarboxylation around the carboxyl groups during the treatment. Since these functional groups are electron-attractive groups, it is expected that the acidity of carboxyl groups becomes stronger in aliphatic acids, as well as even in the case of benzoic acids substituted with these functional groups at the meta position. Also, the acidity becomes strong in the case of benzoic acids substituted with these functional groups at the ortho position because of steric hindrance. In fact, their formation was confirmed in previous study of in situ infrared spectroscopy during pyrolysis of brown coal [27]; however, there is no evidence that these functional groups are present around the carboxyl groups. Because the pKa values of carboxylic acids with this structure are not reported, their acidities were
182
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Fig. 7. Dependence of heat-treatment temperature on the amount of each acidic group (2, Group A; !, Group B; a, Group C).
Fig. 8. The heat-treatment temperature dependence of the wavenumber of the peak of carboxyl groups for Loy Yang brown coal.
3.4. Comparison of acidities obtained by titration method with the FT-IR results Fig. 6. Acidity distributions for heat-treated brown coals.
estimated using the MOPAC program (Hamiltonian: AM1, keyword: PRECISE). This calculation of pKa value was performed by the following procedure [28]; (1) the electron densities on the protons of the carboxyl groups in simple carboxylic acids, which are characterized by known pKa values [25] were calculated by MOPAC program; (2) the linear relationship between these electron densities and their pKa values was established; and finally (3) the pKa values of carboxyl groups were estimated using this relationship. As a result, the calculated acid dissociation constants were 2.7–3.9, and these values are much lower than the pKa values of the simple carboxylic acids. Hence, such carboxyl groups are also likely to be contained in the heat-treated coals.
As described in the above section, only the strongest carboxyl groups were observed to remain on coal samples heat-treated at 400 ◦ C by titration measurement, which is consistent with the results of FT-IR measurement reported in our recent paper [13]. We measured there the change of FT-IR spectra in the same Loy Yang brown coal under heat-treatment until 400 ◦ C. As shown in Fig. 8, the peak attributed to the C=O stretching mode of carboxyl groups remained constant at 1715 cm−1 below 250 ◦ C, but at higher temperatures it shifted to lower wavenumber until it reached 1711 cm−1 at 400 ◦ C. As discussed in the previous paper, it is well known that the carboxyl groups with stronger acidity have a tendency to appear at lower wavenumber.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
4. Conclusions By analyzing the results of aqueous titration assuming Gaussian functions as acidity distribution functions, we can obtain very sharp acidity distributions of some standard carboxylic acids. Further, the number and the acidity of carboxyl groups corresponded with the stoichiometric value and the reference value, respectively. By applying this method to brown coal samples, we examined the change of acidity distribution with heat treatment. Four acidic groups (pKa 3.8, 5.2, 6.8, and 8.3) with different chemical properties were found in the sample dried at 50 ◦ C under vacuum (LY50) and the acidity for each group has a distribution. By heat treatment, the number of carboxyl groups decreased, but the decrease in number was different among acidic groups. At 400 ◦ C, only the strongest acidic groups remained. The background was considered by comparing these experimental pKa values with the pKa values of simple carboxylic acids and calculated by MOPAC.
Acknowledgment This work was financially supported by a Grant-in-Aid for Scientific Research (No. 13750710) from the Ministry of Education, Culture, Science and Technology, Japan.
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
References [1] [2] [3] [4]
H.N.S. Schafer, Fuel 49 (1970) 197. H.N.S. Schafer, Fuel 49 (1970) 271. G.V. Cullen, N.G. Siviour, Water Res. 16 (1982) 1357. A.D. Stuart, Fuel 65 (1986) 1003.
[25] [26] [27] [28]
183
S.J. Allen, Fuel 66 (1987) 1171. C.J. Lafferty, M. Hobday, Fuel 69 (1990) 78. C.J. Lafferty, M.D. Hobday, Fuel 69 (1990) 84. J.H. Kuhr, J.D. Robertson, C.J. Lafferty, A.S. Wong, N.D. Stalnaker, Energy Fuels 11 (1997) 323. C.A. Burns, P.J. Cass, I.H. Harding, R.J. Crawford, Colloids Surf. A 155 (1999) 63. K. Murakami, J. Ozaki, Y. Nishiyama, Fuel Process. Technol. 43 (1995) 95. K. Murakami, T. Yamada, K. Fuda, T. Matsunaga, Y. Nishiyama, Fuel 76 (1997) 1085. K. Murakami, T. Yamada, K. Fuda, T. Matsunaga, Fuel 80 (2001) 599. K. Murakami, T. Yamada, K. Kaga, K. Fuda, T. Matsunaga, Colloids Surf. A 193 (2001) 153. T.J. Bandosz, J. Jagiello, C. Contescu, J.A. Schwarz, Carbon 31 (1993) 1193. J. Jagiello, T.J. Bandosz, J.A. Schwarz, Carbon 32 (1994) 1026. J. Jagiello, T.J. Bandosz, K. Putyera, J.A. Schwarz, J. Colloid Interface Sci. 172 (1995) 341. A. Seron, H. Benaddi, F. Beguin, E. Frackowiak, J.L. Bretelle, M.C. Thiry, T.J. Bandosz, J. Jagiello, J.A. Schwarz, Carbon 34 (1996) 481. T.J. Bandosz, J. Jagiello, J.A. Schwarz, Langmuir 12 (1996) 6480. T.J. Bandosz, B. Buczek, T. Grzybek, J. Jagiello, Fuel 76 (1997) 1409. A. Contescu, C. Contescu, K. Putyera, J.A. Schwarz, Carbon 35 (1997) 83. H. Benaddi, T.J. Bandosz, J. Jagiello, J.A. Schwarz, J.N. Rouzaud, D. Legras, F. Beguin, Carbon 38 (2000) 669. C. Contescu, V.T. Popa, J.A. Schwarz, J. Colloid Interface Sci. 180 (1996) 149. C. Contescu, V.T. Popa, J.B. Miller, E.I. Ko, J.A. Schwarz, J. Catal. 157 (1995) 244. A.M. Puziy, O.I. Poddubnaya, J.A. Ritter, A.D. Ebner, C.E. Holland, Carbon 39 (2001) 2313. A.E. Martell, R.M. Smith, Critical Stability Constants, Vols. 1, 2, 3, 5, Plenum, New York, 1974, 1975, 1977, 1982. K. Murakami, H. Shirato, J. Ozaki, Y. Nishiyama, Fuel Process. Technol. 46 (1996) 183. K. Murakami, H. Shirato, Y. Nishiyama, Fuel 76 (1997) 655. M.J. Citra, Chemosphere 38 (1999) 191.
Acidity distribution of carboxyl groups in Loy Yang brown coal: its analysis and the change by heat treatment Kenji Murakami,∗ Ryuhei Kondo, Kiyoshi Fuda, and Toshiaki Matsunaga Department of Materials-Process Engineering & Applied Chemistry for Environments, Faculty of Engineering and Resource Science, Akita University, Akita 010-8502, Japan Received 30 July 2002; accepted 24 November 2002
Abstract Brown coals have a considerable number of acidic functional groups of which the main component is carboxyl groups, and the acidity has a wide distribution. In this paper, changes of the acidity distribution were examined by aqueous titration when brown coal was heat-treated to control its acidity distribution. For Loy Yang brown coal from Australia dried at 50 ◦ C under vacuum (LY50), the acid dissociation constant, Ka, was distributed over a wide pKa range between 2 and 9. Then, using Gaussian functions, the acidity distribution was divided into four groups, which were characterized by average pKa values: average pKa value of 3.8 (hereafter referred to as Group A), 5.2 (Group B), 6.8 (Group C), and 8.3 (Group D). Among them, Groups A, B, and C were assigned to carboxyl groups. From the changes of the number of carboxyl groups when brown coal was heat-treated up to 400 ◦ C, it was found that the way of decrease was different among these acidic groups. The decrease of the amount of carboxyl groups in Group C was significant, and at 325 ◦ C most of them disappeared. On the other hand, the carboxyl groups in Group A remained even at a high temperature of 400 ◦ C. We estimated approximately the structures around carboxyl groups for LY50 and their structural changes by heat treatment using the known pKa values for simple carboxylic acids and the pKa values calculated by the MOPAC program for complicated carboxylic acids. 2003 Elsevier Science (USA). All rights reserved. Keywords: Brown coal; Carboxyl group; Acidity distribution; Heat treatment; Titration; MOPAC calculation
1. Introduction It is well known that brown coals have the ability of cation exchange and carboxyl groups play an important role [1–13]. Individual carboxyl group existing in brown coal should have different acidity due to the variety of environmental conditions, including the substrates bound with carboxyl groups. In the field of catalytic study, analytical methods such as NH3 -TPD and IR study of adsorbed amines have been used to investigate the acidity of solid acids. However, the surface acidity measured by such solid–gas reactions should be considerably different from that measured by solid–liquid reactions (e.g., the adsorption of metallic ions onto solid acids in aqueous solutions). We examined the acidity distribution of carboxyl groups on the surface of Australian Loy Yang brown coal from the results of aqueous titration in our earlier report [11]. However, at that time we * Corresponding author.
E-mail address: [email protected] (K. Murakami).
obtained only a small number of experimental data; therefore we could obtain only rough acid-dissociating properties of carboxyl groups. Recently, by analyzing a proton-binding isotherm (this isotherm represents the change of the protonbinding number of acidic functional groups as a function of the pH of the solution obtained by aqueous titration, and this term is used by Schwarz et al.), the surface acidities of carbonaceous materials such as activated carbons [14–21] and other materials such as titania [22], mixed oxides (TiO2 –SiO2 , ZrO2 –SiO2 ) [23], and carbon composite adsorbents (oxidized carbon and titanium silicate) [24] have been evaluated. According to these reports, it was clear that there were some acidic functional groups with different acidities on the surfaces of these heterogeneous materials. In this study, we made automatic titration equipment which can drop small amounts of reagent to increase the number of experimental data and to analyze the titration data in detail. Here, we assumed that the acidity distribution function was Gaussian, similarly to our earlier study [11] and Contescu et al. [20]. As described later, we confirmed that a very sharp acidity distribution could be obtained by ana-
0021-9797/03/$ – see front matter 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S0021-9797(02)00172-8
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
lyzing the acidity distributions of some pure carboxylic acids (from monobasic acid to tribasic acid). The purpose of this study is to clarify the acidity distribution of brown coal and to investigate the change in acidity distribution when brown coal was heat-treated to control its acidity distribution. Also, we estimated the structural changes around carboxyl groups by comparing the pKa values calculated by the MOPAC program for carboxylic acids with unknown pKa values.
2. Experimental 2.1. Sample preparation Loy Yang brown coal from Victoria, Australia was used in this study. Raw coal was ground below particle size 250 µm, washed with deionized water, dried at 50 ◦ C under vacuum, and stored in a desiccator. This sample is a starting material (hereafter referred to as LY50). The analyses for LY50 are as follows: C 67.6% (daf), H 5.2% (daf), N 0.8% (daf), O 26.4% (diff), and ash 0.2% (dry). A sample of 10 g of LY50 was heat-treated at a rate of 10 ◦ C min−1 up to desired temperatures (250–400 ◦ C) under nitrogen flow (50 ml min−1 ) and then held at the temperatures for 2 h (LY250–LY400). To examine the reliability of the experimental and analytical method used in this study, some carboxylic acids (acetic acid, phthalic acid, and citric acid) which are characterized by known acid dissociation constants were used as standard reagents. 2.2. Titration method A weighed amount of sample (0.5 g) was added into a 200-ml Erlenmeyer flask containing 100 ml of 0.1 mol l−1 NaOH solution. The suspension was continuously saturated with nitrogen to eliminate the influence of atmospheric CO2 and stirred magnetically throughout the measurements. The adsorption system was set in a vessel thermostated at 25 ± 0.2 ◦ C. The titrations were carried out with a 0.1 mol l−1 HCl solution using hand-made titration equipment and a pH meter (Horiba F-22) connected to a PC computer, which controlled the equilibrium condition and titrant delivery. The titration equipment is able to dose the titrant in increments of 1 ml to the suspension of samples. The pH of the solution was measured every 10 min. We assumed that the suspension reached equilibrium when the pH variation was less than both 0.01 pH units/10 min and 0.005 pH units/30 min.
expressed as Ka =
[COO− ][H+ ] , [COOH]
In an equilibrium condition, the acid dissociation constant, Ka, of weak acid such as carboxylic acids is generally
(1)
where [COOH], [COO− ], and [H+ ] are the molar concentrations (mol l−1 ) of undissociated carboxyl groups, dissociated carboxyl groups, and protons, respectively. Although fundamentally the acid dissociation constant should use activities instead of molar concentrations, we approximate the acid dissociation constant in this way. As described below, the results of simulations using this approximation for some carboxylic acids with known pKa values were in good correspondence with the values reported in the reference. Therefore, this approximation is considered to be valid. The degree of electrolytic dissociation, α, of carboxyl groups can be expressed in terms of Ka and [H+ ] as α=
[COO− ] Ka [COO− ] , (2) = = [COOH]T [COOH] + [COO− ] [H+ ] + Ka
where [COOH]T is the total molar concentration (mol l−1 ) of carboxylic acid. By differentiating Eq. (2) with respect to pH of the solution, the following equation can be obtained and is named q(pH, pKa): q(pH, pKa) =
loge 10 [H+]Ka . ([H+ ] + Ka)2
(3)
This Eq. (3) represents the number of dissociated carboxyl groups at each pH of the solution. In the case of a heterogeneous solid such as brown coal, the acid dissociation constant of carboxyl groups present on the surface is considered to have a wide distribution. This distribution being named as f (pKa), the dissociated amount of carboxyl groups at each pH, Q(pH), can be written in the form ∞ Q(pH) =
f (pKa)q(pH, pKa) dpKa,
(4)
−∞
where Q(pH) is a measurable amount. By deconvoluting Eq. (4), the acidity distribution, f (pKa), can be obtained. Although we do not know what function should be applied to the acidity distribution function, as described below we could reproduce the experimental values accurately by using a Gaussian function, which is most suitable for probability phenomena: f (pKa) = G(Ni , pKai , σi ) i
=
i
2.3. Analytical method of acidity distribution
177
pH − pKai exp −0.5 √ σi 2π σi Ni
2 .
(5)
In this study, f (pKa) was taken as an overlap of several Gaussian functions with the number of carboxyl groups Ni , the acid dissociation constant pKai , and the standard deviation σi .
178
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
3. Results and discussion 3.1. Examination of reliability of this experimental method by measuring the amount and acidity of carboxyl groups of standard reagents We evaluated the reliability of the experimental and analytical methods used in this study by titrating some pure materials such as acetic acid (monobasic acid), phthalic acid (dibasic acid), and citric acid (tribasic acid) as standard reagents. Figure 1 shows the procedure of analysis in the case of acetic acid as a example. The difference in sodium contents in solution between the blank titration curve and the titration curve with sample at the same pH, x in Fig. 1a, was taken as the amount of sodium binding. The curve showing the number of carboxyl groups bound with sodium ions as a function of the pH of the solution (Fig. 1b) was named the sodium binding isotherm. This isotherm was regarded as a measure of the amount of dissociated carboxyl groups. By differentiating this isotherm, the curve shown in Fig. 1c was
obtained. This differential curve is equivalent to Eq. (3) and named as the acid dissociation curve, Q(pH). The numerical values in the vertical axes (Figs. 1b and 1c) are normalized by the initial concentration of acetic acid. By deconvoluting Eq. (4) using Q(pH) obtained in the experiment, the f (pKa) of acetic acid was obtained (Fig. 1d). Similarly, for phthalic acid and citric acid, the sodium binding isotherms, the acid dissociation curves, and f (pKa) are shown in Figs. 2 and 3. Because we use a Gaussian function for analysis, the pKa value was distributed even in the case of carboxylic acids which should have a single pKa value inherently. But the widths of the acidity distribution were very narrow, as shown in Figs. 1d, 2c, and 3c (σi = 0.05–0.07 pKa units for any samples). For citric acid, these widths were narrower than the widths reported by other researchers, who analyzed by SAIEUS (the solution method for the integral equation using a B-spline function) [16] and CONTIN (a general purpose constrained regularization method) [24]. The analytical results are summarized in Table 1 together with the reference values [25]. All Ni parameters of each standard reagent are
Fig. 1. Procedure for calculation of acidity distribution for acetic acid. (a) Broken line is a blank curve and solid line is a titration curve for acetic acid. x indicates the amount of sodium binding. (b) The sodium binding isotherm. (c) The acid dissociation curve. Symbols, experimental data; solid line, calculated using Eq. (4). (d) The acidity distribution.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Fig. 2. (a) Sodium-binding isotherm for phthalic acid. (b) The acid dissociation curve for phthalic acid. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for phthalic acid.
179
Fig. 3. (a) Sodium-binding isotherm for citric acid. (b) The acid dissociation curve for citric acid. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for citric acid.
180
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Table 1 Calculated Ni , total Ni , and pKai parameters for standard reagents and the pKai values reported in Ref. [25] Ni (mol mol−1 ) Acetic acid 0.99 Phthalic acid 0.91 1.05 Citric acid 0.96 0.91 1.08
Total Ni (mol mol−1 )
pKai measured
pKai∗
0.99
4.62
4.56
1.96
2.74 4.98
2.75 4.93
2.95
3.04 4.38 5.66
2.90 4.34 5.66
pKai∗ : values reported in Ref. [25] (25 ◦ C, ionic strength = 0.1 mol dm−3 ).
about 1 mol l−1 , which are almost equal to their stoichiometric amounts. Further, the pKai parameters of all reagents are also in good agreement with the values reported in the reference [25]. Therefore, it is concluded that this method in our study can be used not only for quantitative but also for qualitative analysis of the titration results. 3.2. Calculation of acidity distribution and estimation of the structure around carboxyl groups for LY50 Figures 4a and 4b show the sodium-binding isotherm and the acid dissociation curve for LY50, respectively. This isotherm was an average of several experiments and the experimental values were in error by up to 4%. From Fig. 4b, the acid dissociation curve for LY50 was found to be very broad. In general, the acid dissociation constants of the carboxyl groups on the surface of carbonaceous materials such as brown coals and activated carbons are well known to have a wide distribution in the pKa range between 3 and 8. The reason the acid dissociation constants are distributed over such a wide range is considered to be the variety of interactions between carboxyl groups and substrates. The results of deconvolution of Eq. (4) are shown in Fig. 4c and Table 2. The obtained parameters (Ni , pKai , σi ) were further refined by fitting to the original sodium binding isotherm. Here, we used 10 Gaussian functions to deconvolute Eq. (4), but it was found from Fig. 4c that LY50 had only four groups of acidic functional groups in the pKa range between 2 and 9. The groups of acidic functional groups are referred to as Groups A, B, C, and D in order of increasing pKa values. The σi values for LY50 (σi = 0.35–0.36 pKa units) were about 5–7 times larger than those for carboxylic acids shown in Figs. 1d, 2c, and 3c, indicating clearly that the acid dissociation constants for LY50 have a wide distribution. We consider that heterogeneous materials with complicated structures such as brown coals, have a continuous pKa distribution over a much broader pKa range, but it was found from Fig. 4c that brown coal was composed of four groups of acidic functional groups which have relatively uniform chemical properties represented at these pKa values.
Fig. 4. (a) Sodium-binding isotherm for LY50. (b) The acid dissociation curve for LY50. Symbols, experimental data; broken lines, calculated using Eq. (4); solid line, sum of each acidic group. (c) Acidity distribution for LY50.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
181
Table 2 Calculated Ni , pKai , and total Ni parameters for coal samples Group A LY50 LY250 LY300 LY325 LY350 LY400
Group B
Group C
Ni
pKai
Ni
pKai
Ni
pKai
0.85 0.67 0.40 0.34 0.23 0.20
3.81 3.79 3.45 3.42 3.44 3.37
0.92 0.71 0.58 0.32 0.15 0
5.19 5.40 4.94 5.33 5.10 –
0.72 0.45 0.24 0 0 0
6.83 6.25 6.29 – – –
Total Ni 2.49 1.83 1.22 0.66 0.38 0.20
What kinds of acidic functional groups are represented at these four pKa values? According to Ref. [25], the acid dissociation constants of aliphatic monocarboxylic acids and benzoic acids are 4.5–4.9 and 4.0–4.6, respectively. On the other hand, the first and second acid dissociation constants of dicarboxylic acids are 1.7–4.5 and 4.1–5.8, respectively. These simple carboxylic acids themselves cannot be present in brown coal, but these pKa values may be useful for obtaining rough information on the structures around carboxyl groups. The average pKa value of Group A is about 3.8. Since this is a strong group, this acidic site would correspond with a structure similar to the benzoic acids and dicarboxylic acids (pK1 ). The average pKa value of Group B is about 5.2, which possibly corresponds with the pKa values of aliphatic monocarboxylic acids and the pK2 values of dicarboxylic acids. The average pKa value of Group C is about 6.8. For these carboxyl groups, we must consider the third acid dissociation constants where many carboxyl groups are present locally. The average pKa value of Group D is about 8.3, which is much higher than the pKa values of the simple carboxylic acids [25]. However, when we carried out the titration experiment for the weakly acidic cation-exchange resin Amberlite IRC-50, its acid dissociation constant was distributed over a wide pKa range, between 4 and 10. Consequently, there is a possibility that this Group D is also assigned to the carboxyl groups. On the other hand, Puziy et al. reported that the acidic site with pKa 7.7–8.0 might be assigned to phenolic hydroxyl groups [24]. In any case, since the acidic group with such higher pKa values is not actually concerned with the ion-exchange reaction, Group D is omitted from the following discussion. 3.3. Influence of heat treatment on the acidity distribution of carboxyl groups The sodium binding isotherms for the heat-treated coals are shown in Fig. 5. The acidity distribution functions for heat-treated coals were calculated similarly to the case of LY50, and results are shown in Fig. 6 and Table 2. For all samples, 10 Gaussian functions were used in deconvolution, but only four peaks appeared at the analogous pH values for LY50. Figure 7 shows the dependence of the number of carboxyl groups for each group on the heat-treatment temperature. From previous studies [26,27], it was clear that CO2 was produced by the decomposition of carboxyl
Fig. 5. Sodium-binding isotherms for heat-treated brown coals.
groups; therefore the number of carboxyl groups decreased with heat treatment. However, the way of decrease in number was different among acidic groups. Group A, which is the strongest acidic group, was difficult to decompose and 0.2 mmol g−1 of carboxyl groups remained, even at 400 ◦ C. On the other hand, the number of Group C decreased significantly and most of them disappeared at 325 ◦ C. Also, most of the carboxyl groups which belonged to Group B disappeared at 400 ◦ C. The average pKa values of each group are shown in Table 2. The average pKa value of Group A shifted from 3.8 to 3.4 gradually with increasing heat-treatment temperature. The average pKa value of Group B was not changed regularly and scattered at the center of about 5.2. From these results, we estimated the structural changes around carboxyl groups with heat treatment. From the fact that Group B and Group C were easy to decompose and that Group A was difficult, the more strongly acidic carboxyl groups were considered to remain without decomposition. But it is not necessarily concluded that only this decomposition behavior occurs, and there may be another possibility, that the pKa values of carboxyl groups which belonged to Group B or Group C shift to the pKa values of Group A with heat treatment. This can occur if carbonyl or ether groups are formed by dehydration or decarboxylation around the carboxyl groups during the treatment. Since these functional groups are electron-attractive groups, it is expected that the acidity of carboxyl groups becomes stronger in aliphatic acids, as well as even in the case of benzoic acids substituted with these functional groups at the meta position. Also, the acidity becomes strong in the case of benzoic acids substituted with these functional groups at the ortho position because of steric hindrance. In fact, their formation was confirmed in previous study of in situ infrared spectroscopy during pyrolysis of brown coal [27]; however, there is no evidence that these functional groups are present around the carboxyl groups. Because the pKa values of carboxylic acids with this structure are not reported, their acidities were
182
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
Fig. 7. Dependence of heat-treatment temperature on the amount of each acidic group (2, Group A; !, Group B; a, Group C).
Fig. 8. The heat-treatment temperature dependence of the wavenumber of the peak of carboxyl groups for Loy Yang brown coal.
3.4. Comparison of acidities obtained by titration method with the FT-IR results Fig. 6. Acidity distributions for heat-treated brown coals.
estimated using the MOPAC program (Hamiltonian: AM1, keyword: PRECISE). This calculation of pKa value was performed by the following procedure [28]; (1) the electron densities on the protons of the carboxyl groups in simple carboxylic acids, which are characterized by known pKa values [25] were calculated by MOPAC program; (2) the linear relationship between these electron densities and their pKa values was established; and finally (3) the pKa values of carboxyl groups were estimated using this relationship. As a result, the calculated acid dissociation constants were 2.7–3.9, and these values are much lower than the pKa values of the simple carboxylic acids. Hence, such carboxyl groups are also likely to be contained in the heat-treated coals.
As described in the above section, only the strongest carboxyl groups were observed to remain on coal samples heat-treated at 400 ◦ C by titration measurement, which is consistent with the results of FT-IR measurement reported in our recent paper [13]. We measured there the change of FT-IR spectra in the same Loy Yang brown coal under heat-treatment until 400 ◦ C. As shown in Fig. 8, the peak attributed to the C=O stretching mode of carboxyl groups remained constant at 1715 cm−1 below 250 ◦ C, but at higher temperatures it shifted to lower wavenumber until it reached 1711 cm−1 at 400 ◦ C. As discussed in the previous paper, it is well known that the carboxyl groups with stronger acidity have a tendency to appear at lower wavenumber.
K. Murakami et al. / Journal of Colloid and Interface Science 260 (2003) 176–183
4. Conclusions By analyzing the results of aqueous titration assuming Gaussian functions as acidity distribution functions, we can obtain very sharp acidity distributions of some standard carboxylic acids. Further, the number and the acidity of carboxyl groups corresponded with the stoichiometric value and the reference value, respectively. By applying this method to brown coal samples, we examined the change of acidity distribution with heat treatment. Four acidic groups (pKa 3.8, 5.2, 6.8, and 8.3) with different chemical properties were found in the sample dried at 50 ◦ C under vacuum (LY50) and the acidity for each group has a distribution. By heat treatment, the number of carboxyl groups decreased, but the decrease in number was different among acidic groups. At 400 ◦ C, only the strongest acidic groups remained. The background was considered by comparing these experimental pKa values with the pKa values of simple carboxylic acids and calculated by MOPAC.
Acknowledgment This work was financially supported by a Grant-in-Aid for Scientific Research (No. 13750710) from the Ministry of Education, Culture, Science and Technology, Japan.
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
References [1] [2] [3] [4]
H.N.S. Schafer, Fuel 49 (1970) 197. H.N.S. Schafer, Fuel 49 (1970) 271. G.V. Cullen, N.G. Siviour, Water Res. 16 (1982) 1357. A.D. Stuart, Fuel 65 (1986) 1003.
[25] [26] [27] [28]
183
S.J. Allen, Fuel 66 (1987) 1171. C.J. Lafferty, M. Hobday, Fuel 69 (1990) 78. C.J. Lafferty, M.D. Hobday, Fuel 69 (1990) 84. J.H. Kuhr, J.D. Robertson, C.J. Lafferty, A.S. Wong, N.D. Stalnaker, Energy Fuels 11 (1997) 323. C.A. Burns, P.J. Cass, I.H. Harding, R.J. Crawford, Colloids Surf. A 155 (1999) 63. K. Murakami, J. Ozaki, Y. Nishiyama, Fuel Process. Technol. 43 (1995) 95. K. Murakami, T. Yamada, K. Fuda, T. Matsunaga, Y. Nishiyama, Fuel 76 (1997) 1085. K. Murakami, T. Yamada, K. Fuda, T. Matsunaga, Fuel 80 (2001) 599. K. Murakami, T. Yamada, K. Kaga, K. Fuda, T. Matsunaga, Colloids Surf. A 193 (2001) 153. T.J. Bandosz, J. Jagiello, C. Contescu, J.A. Schwarz, Carbon 31 (1993) 1193. J. Jagiello, T.J. Bandosz, J.A. Schwarz, Carbon 32 (1994) 1026. J. Jagiello, T.J. Bandosz, K. Putyera, J.A. Schwarz, J. Colloid Interface Sci. 172 (1995) 341. A. Seron, H. Benaddi, F. Beguin, E. Frackowiak, J.L. Bretelle, M.C. Thiry, T.J. Bandosz, J. Jagiello, J.A. Schwarz, Carbon 34 (1996) 481. T.J. Bandosz, J. Jagiello, J.A. Schwarz, Langmuir 12 (1996) 6480. T.J. Bandosz, B. Buczek, T. Grzybek, J. Jagiello, Fuel 76 (1997) 1409. A. Contescu, C. Contescu, K. Putyera, J.A. Schwarz, Carbon 35 (1997) 83. H. Benaddi, T.J. Bandosz, J. Jagiello, J.A. Schwarz, J.N. Rouzaud, D. Legras, F. Beguin, Carbon 38 (2000) 669. C. Contescu, V.T. Popa, J.A. Schwarz, J. Colloid Interface Sci. 180 (1996) 149. C. Contescu, V.T. Popa, J.B. Miller, E.I. Ko, J.A. Schwarz, J. Catal. 157 (1995) 244. A.M. Puziy, O.I. Poddubnaya, J.A. Ritter, A.D. Ebner, C.E. Holland, Carbon 39 (2001) 2313. A.E. Martell, R.M. Smith, Critical Stability Constants, Vols. 1, 2, 3, 5, Plenum, New York, 1974, 1975, 1977, 1982. K. Murakami, H. Shirato, J. Ozaki, Y. Nishiyama, Fuel Process. Technol. 46 (1996) 183. K. Murakami, H. Shirato, Y. Nishiyama, Fuel 76 (1997) 655. M.J. Citra, Chemosphere 38 (1999) 191.